Can machines surpass humans in intelligence? People were shocked in 1997 when IBM’s Deep Blue computer beat Garry Kasparov, a Russian grandmaster, at chess. But winning a board game is a trivial task compared with understanding the complexities and idiosyncrasies of human speech. The company has now developed Watson, a supercomputer it thinks is capable of understanding “natural language”.

To put this claim to the test, IBM arranged for its creation to compete in Jeopardy!, an American quiz show known for using clues and wordplay that even bright humans struggle to understand. In the contest, televised in America in February, Watson trounced the two most successful previous champions of the quiz.

This victory fits nicely into Ray Kurzweil’s vision. An inventor and futurist, he has long predicted the rise of intelligent machines. Transcendent Man, a new documentary film, probes his breathtaking, possibly balmy, vision of the future.Mr Kurzweil leads an influential cabal of techno-optimists, a group that includes Silicon Valley stars, scientific grandees and even the Obama administration’s chief information officer, Vivek Kundra. They believe mankind is heading for a glorious post-biological era known as the Singularity. Thanks to implants and other enhancements, humans will improve along with machines. But artificial intelligence will inevitably surpass the human kind — and will do so, according to Mr Kurzweil’s calculations, as early as 2029.

Predicting what will happen after that point is difficult, he argues, because we cannot hope to predict the behaviour or evolution of hyper-intelligent machines. But he insists that “the intelligence that will emerge will continue to represent the human civilisation, which is already a human-machine civilisation.” […]